At the recent Google I/O 2021 conference, the cloud provider announced the general availability of Vertex AI, a managed machine learning platform designed to accelerate the deployment and maintenance of artificial intelligence models.

Using Vertex AI, engineers can manage image, video, text, and tabular datasets, build machine learning pipelines to train and evaluate models using Google Cloud algorithms or custom training code. They can then deploy models for online or batch use cases all on scalable managed infrastructure.

The new service provides Docker images that developers run for serving predictions from trained model artifacts, with prebuilt containers for TensorFlow, XGBoost and Scikit-learn prediction. If data needs to stay on-site or on a device, Vertex ML Edge Manager, currently experimental, can deploy and monitor models on the edge.

Vertex AI replaces legacy services such as AI Platform Data Labeling, AI Platform Training and Prediction, AutoML Natural Language, AutoML Video, AutoML Vision, AutoML Tables, and AI Platform Deep Learning Containers.

Andrew Moore, vice president and general manager of Cloud AI at Google Cloud, explains why the cloud provider decided to introduce a new platform:

Cassie Kozyrkov, chief decision scientist at Google, highlights the main benefit of the new product, managing the entire lifecycle of AI and machine learning development

In one of the comments, Ornela Bardhi, Marie Curie PhD fellow in AI and health at the University of Deusto, praises the new service but raises a question about accountability of managed services in machine learning:

Some users on Reddit question instead if the announced platform is simply a rebranding, as user 0xnld suggests:

In a separate article, Google explains how to streamline ML training workflows with Vertex AI, avoiding running model training on local environments like notebook computers or desktops and working instead with Vertex AI custom training service. Using a pre-built TensorFlow 2 image as example, the authors cover how to package the code for a training job, submit a training job, configure which machines to use and access the trained model.

The pricing model of Vertex AI matches the existing ML products that it will supersede. For AutoML models, users pay for training the model, deploying it to an endpoint and using it to make predictions.

Microsoft has made available TypeScript 4.3, the latest release of its programming language that builds on JavaScript and follows the release candidate that was made available earlier this month.

New in this release is support for separate write types on properties, to address the issue of APIs converting values that are passed in before storing them if the developer has not explicitly defined what type a value is. As of 4.3, users can specify types for reading and writing to properties.

TypeScript 4.3 has also introduced the override keyword to use when extending classes. This is to avoid the unintended consequences if a developer does not make it explicit whether they meant to add a new method or override an existing one.

Breaking changes in this release stem from the removal of APIs that no browser implements in declarations generated for web contexts, while if a value with a union enum type is compared with a numeric literal that it could never be equal to, then the type-checker will now issue an error. See the Microsoft DevBlog for further details.

Red Hat provides Universal Base Images on Docker Hub
Advertisement
Red Hat has rolled out Red Hat Universal Base Images on Docker Hub, claiming this will make it easier for organisations to build cloud-native applications based on tested and trusted components of Red Hat Enterprise Linux.

Red Hat Universal Base Images (UBI) are containerised operating system images that include complementary runtime languages and packages. They are Open Container Initiative (OCI)-compliant and freely redistributable, the firm said.

UBI images are offered in a variety of configurations tailored for different enterprise IT use cases. These include Standard, which provides the necessary runtimes and YUM repositories to build, deploy and share UBI-based containers, while Minimal contains the bare essentials for a lightweight RHEL-based image.

Previously, UBI could only be obtained from the Red Hat container catalogue. With availability extended to Docker Hub, it will be easier for UBI-based containers to be built and deployed anywhere, Red Hat claimed.

Databricks AutoML makes it easier to build machine learning models
DataBricks has announced a public preview of Databricks AutoML, a tool designed to help users build and deploy machine learning models by automating the heavy lifting of pre-processing, feature engineering and model training/tuning.

According to DataBricks, AutoML would allow users to select a dataset, then train and deploy a model entirely through a UI. It integrates with the Databricks ML ecosystem, including tracking trial run metrics and parameters with MLflow, allowing users to register and version control models in the Databricks Model Registry for deployment.

AutoML provides Python notebooks for every model, with the source code for each trial run so users can review, reproduce, and modify the code. Data scientists can use their expertise to add or modify cells to these generated notebooks, and use them to jumpstart ML development by bypassing the need to write boilerplate code.

GitHub champions OpenTelemetry
GitHub has announced it is working to adopt the CNCF OpenTelemetry project internally, as a way to improve observability of its own infrastructure. The GitHub team said the project has potential to transform how the entire industry approaches ways to observe and understand systems, and invited other developers to provide assistance.

OpenTelemetry introduces a common, vendor-neutral format for telemetry signals, called OpenTelemetry Protocol (OTLP). It also enables telemetry signals to be easily correlated with each other. GitHub said OpenTelemetry would enable it to build integrated and opinionated solutions for its engineers, so that designing for observability would be at the forefront of its engineer’s minds.

GitHub said that as its OpenTelemetry efforts progress, it intends to shift focus towards logging and metrics systems, and claimed it is contributing as much as possible back to the OpenTelemetry project for others to benefit from.

Digital.ai intros platform to improve Devops outcomes
Digital.ai has announced the Digital.ai Platform, which uses machine learning-based analytics in a bid to predict and prevent issues impacting software delivery and reliability.

The platform features end-to-end DevOps lifecycle orchestration capabilities, such as release management, risk management, and software delivery predictability. It has strong integration and collaboration with DevOps tools, including Azure DevOps, Chef, Jenkins, Atlassian Jira, ServiceNow and Selenium, according to Digital.ai.

CDK for Terraform 0.4 gets Go
HashiCorp has released CDK for Terraform 0.4, the latest update of the Cloud Development Kit for its Terraform infrastructure-as-code provisioning tool.

CDK for Terraform give developers the ability to write Terraform configurations in C#, Python, TypeScript, and Java, using all existing Terraform providers and Terraform modules. This release has added experimental Go support, one of the top requests from users, according to HashiCorp.

Also added is a new asset construct that makes it easy to reference files and folders that need to be deployed alongside newly provisioned resources such as applications, plus improved Terraform Cloud integration. The CDK for Terraform changelog contains a comprehensive list of enhancements and bug fixes applying to this release.

When it comes to new AI automation, the Software 2.0 movement can be seen as akin to the auto industry’s recent evolution. Beginning in the 1990s and rapidly progressing into the 2000s, the auto industry saw a huge rise in AI automation such as digital vehicle diagnostics.

Throughout this, people worried that digital transformation might take jobs away, when instead it led to “greater profits, productivity, and competitiveness,” according to a 2008 study by the journal of Technological Forecasting and Social Change. In short, these innovations didn’t replace jobs, per se; they simply changed the processes of jobs from the manufacturer’s assembly line down to the mechanic’s garage.

An auto diagnostic specialist used to perform tedious tasks such as counting flashes and converting them to error codes on printed-out tables. Computerized diagnostic tools made this process faster, more reliable and, you might imagine, far less tedious. According to diagnostic specialists Helmut Frank and Uwe Schmidts, digital transformation went from being “a necessary evil to being a key to new, interesting and innovative functions.”

You might liken the above example to the innovations that AI automation and machine learning (ML) have offered to software developers, who until recently, were tasked with a much more hands-on approach to deploying and maintaining applications. In many ways, DevOps practices have heroically come in to break them free of this minutiae, automating many of the repetitive tasks involved in application management. By design, it’s increasing efficiency and security while decreasing tedium and the chance for human error.

Speaking to tedium, Mike Loukides and Ben Lorica, in their article “The road to Software 2.0” noted, “Up until now, we’ve built systems by carefully and painstakingly telling systems exactly what to do, instruction by instruction. The process is slow, tedious, and error-prone; most of us have spent days staring at a program that should work, but doesn’t.”

In short: Much like digital transformation brought efficiency to the life cycle of automobiles, AI automation is offering efficiency to every stage of the software development cycle.

Don’t Change the Players, Change the Game
Indeed, a new generation of AI automation and ML tools is now emerging, shaping what some call a Software 2.0 movement. But before you begin to fear that these tools will lead to a “they took our jobs!” revolt, consider that they may be celebrated instead as a means of not only augmenting the work of software developers but also helping offset the ongoing developer drought. According to a 2019 Stack Overflow survey, over a third of developers named “not enough people for the workload” as their primary challenge to productivity. Rather than replacing jobs, automation should be viewed as offering efficiency in an industry that desperately needs it.

As an example, the process of writing code, though more streamlined today than it was 25 years ago, has remained a mostly manual process. This has presented an excellent opportunity for the automated processes we are beginning to see, which bring in AI to automate aspects of code development such as program synthesis, smart code completion and static code analysis.

Assaf Araki spoke to this in the article, “Software 2.0 takes shape.” Araki described how automation is filling gaps in many other areas, such as regression testing, code analysis, AI for vulnerability detection and observability tools to monitor software health.

From Monolith to Microservices
Just like the recent, mysterious disappearing monolith in the Utah desert, the relevancy of monolithic development may be on its way out.

In the past, software cycles were longer. Teams updated monolithic-style deployments only monthly or yearly. Nowadays, companies are under pressure to push much more rapidly, often turning to microservices and progressive release strategies. This new paradigm requires much automation to increase release efficiencies and enable a seamless CI/CD.

As Araki pointed out, “Automation in software is best viewed through the lens of Continuous Integration (CI) and Continuous Deployment (CD).”

How to Know When AI Automation is Worth Its Salt
It’s smart to consider the investment of both money and time when deciding whether an AI solution is worth adopting. “AI systems need to add value that exceed any new costs and requirements (such as the need for labeled data, hiring new staff, etc.),” Araki said.

You should also be asking if the solutions are wide-ranging enough to meet the needs of your developers and anyone else on the team who may be employing them. Key considerations for AI automation, Araki noted, are focusing on tasks and systems that are frequently used, adjusted or tuned.

4 Examples of Versatile AI Automation and ML:
Data management systems from simple systems that employ ML to more complex systems that enable self-driving data management.
Dashboards that are automatically created, ideally with intuitive and visually appealing UX, that use AI automation and ML to bring up the right elements—more progress is expected in these areas in the coming years.
Websites and mobile apps that use AI automation to become highly personalized interfaces, such as those that employ real-time data for online and reinforcement learning—these are also expected to progress in the coming years.
UX and design tools that use simple automation to design and revise digital media, with ML being a promising way to kick the automation up a notch, such as deep learning for game developers.
AI Automation and ML to Grant Opportunities
In short, Software 2.0 doesn’t currently appear poised to threaten jobs or decrease developer need. Instead, it seems to be changing the processes engineers are using and creating efficiencies in their work, at a time when their expertise is in increasing demand. Demand for novel roles, like the Cloud Economist, demonstrates the emergence of new opportunities for humans with cloud knowledge who work alongside AI.

The new evolution of AI and ML also could create a unique subset of developers with skills keeping par with these new innovations—much like digital transformation in the auto industry led to evolving specialties in computerized vehicle systems.

Not to mention that just as low-code has paved the way for citizen developers and AutoML led to citizen data scientists to fill large gaps in demand, AI and ML solutions have the potential to further broaden and democratize software and code development.